Sanskar’s Whiteboard

Research notes on systems, artificial intelligence, interpretability, applied mathematics, and ideas worth exploring.

Welcome to Sanskar’s Whiteboard, a technical blog where I explore research ideas and system designs at the intersection of:

Machine learning and AI
Systems architecture
Behavioral analysis of software
Applied mathematics in computing
AI-driven cybersecurity
Interpretability and model understanding

The goal of this blog is to document technical thoughts, experimental ideas, research notes, and conceptual architectures, some of which may later develop into formal research papers.

Why This Blog Exists

Many interesting ideas start as rough sketches on a whiteboard.

This blog serves as a place to:

Think through technical problems in public
Share open research ideas
Write technical explorations and system designs
Connect with researchers and engineers with similar interests

Some posts may present exploratory ideas, while others may take the form of research notes or technical writeups.

Topics Covered

Posts on this blog may explore topics such as:

Machine learning and artificial intelligence systems
Interpretability and understanding learned representations
Behavioral analysis of software and malware
Program analysis and operating system telemetry
Graph and trajectory-based modeling of complex systems
Topological and geometric approaches to data analysis and AI
System architecture and learning dynamics

Many posts focus on conceptual frameworks and system architectures, sometimes accompanied by technical notes, preprints, or early-stage research ideas.

Current Series

Topology-Based Behavioral Malware Detection

An exploration of a conceptual architecture for detecting malware using:

Behavioral trajectories
Manifold embeddings
Persistent homology
Predictive trajectory modeling
Reinforcement learning with human feedback

The central idea:

Malware can mutate its code, but it is much harder to mutate the geometry of its behavior.

Posts in this series

Topology-Based Behavioral Malware Detection (Part 2): Behavioral Trajectories — March 14, 2026
Topology-Based Behavioral Malware Detection (Part 1): From Signatures to Behavioral Geometry — March 10, 2026

Research Writings

Technical essays, research notes, and conceptual explorations published on this blog.

Topology-Based Behavioral Malware Detection (Part 2): Behavioral Trajectories — March 14, 2026
Topology-Based Behavioral Malware Detection (Part 1): From Signatures to Behavioral Geometry — March 10, 2026

These posts often explore ideas that may later develop into technical reports, research papers, or experimental systems.

Structure of the Blog

Posts are typically written as short technical essays or multi-part series exploring a specific idea.

Example structure:

Problem framing
Key conceptual insight
Mathematical or architectural approach
Potential advantages and limitations
Open questions

Open Science

All ideas shared here are intended in the spirit of open scientific exploration.

If you find something interesting, build on it, question it, or reach out to discuss it.

About Me

I’m interested in exploring problems at the intersection of:

interpretability
AI-driven security
machine learning and artificial intelligence
systems design
applied mathematics

This blog serves as a public notebook for those explorations.

Research Directions

Current research directions I’m exploring include:

Geometric and topological approaches to modeling software and system behavior
Behavioral trajectory modeling for malware detection and security systems
Structural learning systems where model architecture emerges from constraints or invariants
Alternative inference layers and mechanisms for large language models
Interpretability and understanding learned representations in AI systems
Theoretical perspectives on machine learning and AI systems

Preprints & Technical Notes

A selection of technical notes and preprints associated with ideas discussed on this blog.

Invariant-Driven Structural Learning for Language Modeling (2026)
Zenodo Record | PDF

Additional technical notes and preprints will be added as new work develops.

Citation

If you reference ideas from this blog, please cite the relevant post, writeup, or associated preprint.

General citation format:

Pandey, Sanskar. “Title of Post or Writeup.”
Sanskar’s Whiteboard, Year.
Available at: https://skar07.github.io/…

When a formal preprint is available, please cite the Zenodo version linked in the post.

BibTeX template:

@misc{pandey_whiteboard, author = {Pandey, Sanskar}, title = {Title of Post or Writeup}, year = {Year}, howpublished = {\url{https://skar07.github.io}}, url = {https://skar07.github.io/…} }

Connect

If you find these ideas interesting or would like to collaborate on related problems, feel free to reach out at skarsie841@gmail.com.

You can also open an issue on this repository to discuss ideas, ask questions, or suggest directions for exploration.

License

Content in this repository is shared under an open license unless otherwise specified.